clang-format: Failed to download resource clang-format_bottle_manifest about linuxbrew-core

[joelberkeley]

i'm also seeing the same for brew install idris2, specifically

==> Downloading https://ghcr.io/v2/linuxbrew/core/idris2/manifests/0.4.0_1
curl: (22) The requested URL returned error: 404                              

Error: Failed to download resource "idris2_bottle_manifest"
Download failed: https://ghcr.io/v2/linuxbrew/core/idris2/manifests/0.4.0_1

with

$ brew update
Already up-to-date.
$ brew doctor
Your system is ready to brew.

[joelberkeley]

also similar to #23713 do we know if these are the same problem?

[iMichka]

Fixed by 3726a13

[brettcannon]

I'm seeing the same issue for python-launcher.

[joelberkeley]

does anyone have any idea how to fix this kind of issue? I'd be happy to try but I've no idea where to start

[brettcannon]

@joelberkeley My guess is some GitHub Action workflow that generates that manifest file failed (either due to a bug in the workflow or a hiccup on GitHub's side, e.g. GHR having problems at that exact moment). Probably the best option is to try and find the GitHub Action which creates the manifest files and then try to find the workflow run that was supposed to create the manifest file and failed (or find out in fact that workflow run never existed to begin with).

I found Homebrew/homebrew-core#80982 as the PR on homebrew-core which I assume is what Linuxbrew picked up.

[branchvincent]

i'm not sure of the expected behavior here, but it's trying to pull the bottle from https://ghcr.io/v2/linuxbrew whereas it only exists right now at https://ghcr.io/v2/homebrew.

as homebrew-core is periodically merged into linuxbrew-core, it looks like the homebrew-built bottle is meant to be manually removed so it can be rebuilt here. doesn't look like that happened with these formulae (here is the failure for clang-format)

[joelberkeley]

i'll be honest I've just found a workaround and after a week or so of installation problems I want a break from debugging. I may come back to this, at least in part because it's faster than my current approach, but I don't know when